Search Google for "What car should I buy?" and you'll get over 7 billion results.

Yet, with all this information, using the Internet to make shopping decisions has only gotten more time-consuming and frustrating.

Internet tools like Google and ChatGPT are built from the ground up to simplify access to vast amounts of information in a universally common way.  No matter who you are, if you ask the same question, you’ll get the same answer, personalization is an afterthought - that's why you're still seeing ads for the air fryer you bought a month ago.

So what do we do?  The best (and only) way we have is to battle through the noise for those people who’ve already done the work, and made the same decision — reviewers, message boards and influencers.

And yet, we still end up spending 3 months researching a 3-day vacation.

It's not enough just to have access to information - because if you think about it, we really don’t know who is providing that information. 

• Are they trustworthy? 
• Qualified? 
• Being paid?
• Care about the same things as me? 

One thing you do know is that the more similar that reviewer/advisor/influencer is to you, the more likely the advice they are providing is fitting and helpful.

In the end, what we’re all really searching for - and what we need - are the right people to listen to in order to make the best buying decision.  That's how people are built to make decisions.

And that's how and why we built Sheperd - a sophisticated AI-powered app that provides more fitting and trustworthy buying recommendations based on the decisions and satisfaction of people just like you.

Invest in making your life easier - invest in Sheperd

Why it is a good investment.

It’s a tool for any shopping occasion.

We've identified at least 18 verticals where Sheperd can have a massive impact, including auto, travel, home buying, electronics and retail. This means you’ll have a one stop shop for every decision.

It’s getting traction

• Amazing user activation: Over 70% of visitors became active users
• Incredible engagement: Those users, on average, spent 8 minutes onsite answering 25 questions
• Unbelievable Super Users: 10% of visitors spent, on average, 20 minutes answering 150
• Retention leading indicators: 1​5% of users committed to a recurring relationship with Sheperd - of which - 60% chose to come back 5 minutes a day every day

It’s a big market

We'll monetize our experience by integrating into the existing affiliate market which already spends over $100 Billion acquiring customers

It’s defensible

• Network Effects: Sheperd is designed to take advantage of both data and user network effects. The cost of switching platforms is prohibitively high, and ultimately the platform with the most users/data will be able to have the best models and thus deliver the most value.
• Entry Friction: Incumbent competition will be unable to duplicate our value prop or trusting relationship with the consumer; their business model (advertising) incents them to favor their customers (clients) over users.

It’s run by a great team

Before starting Sheperd, Fede and Mike had worked closely together for 6 years at Ordergroove, a SaaS subscription software company in New York. They even share a few patents related to AI systems born out of projects they spearheaded there.

Additional Information

The Deck

What's been built so far?

Live prototype: https://www.sheperd.com

How will the AI work?

While we can't give away all our secrets, we can cover this at a high level. We've spent the last year and a half researching the potential of our approach, speaking to dozens of experts across many cognitive, behavioral and data science fields.

First, we use existing computational models and assessments to build a decision preference profile for a user. This provides a baseline for our recommendations which will far outperform existing systems on its own.

We then combine data from existing Sheperd users and our own proprietary research to create machine-learning models that tune these profiles - this is the real secret sauce. As our dataset grows we'll be able to continuously optimize the data collection process by asking fewer more 'information-packed' questions that also provide insights across decision verticals. As a high level example, our existing data has already uncovered promising correlations between answers - see the image below

How will we make money?

There is an existing affiliate ecosystem that is long established where over $100 billion is spent in the US per year on acquiring customers.

What impact does ChaptGPT have on Sheperd?

The media frenzy around ChatGPT has had an amazingly positive impact on our prospects. In particular:

• It has shattered the long-held assumption that Google search is invulnerable.
• Its inability to answer subjective questions has highlighted its and Google’s inability to provide users with the same personalized information that’s available from Sheperd
• They’ve increased the value of our data. No generative AI built on public data will be defensible - only AI with highly proprietary and complementary data will stand out. There isn't a product in existence better positioned to win that arms race than Sheperd.

Quotes from a conversation with ChatGPT:

... it then suggested a number of models ...

What kind of research have we done?

We've interviewed car dealerships from all over the country to learn about how they use leads and how they make decisions on lead vendors: